Damage detection in refrigerator compressors using vibration and current signatures

The compressor is an important subsystem of the refrigerator that drives the reliability of the whole system. The present work addresses effective methods to quantify the accumulated mechanical damage early in the design cycle. A sample population is tested under accelerated stress conditions for a limited time. The pressure differential between the discharge and the suction sides, lubricant viscosity, temperature, and number of on/off cycles are used as stress variables. System degradation is monitored through several damage metrics and trended over time. Statistical models are built to describe the damage metrics obtained by experiment and to compote confidence bounds. The present method requires neither testing systems to failure nor is dependent on any time to failure data (although any failure data can be incorporated). It is rather based on the signature of the developing degradation mechanisms that take place under loads in time. The proposed degradation metrics can be used to compare the quality of different refrigerator compressor suppliers, on a relative basis, and/or can be used as a predictive tool for system performance in time. The dynamic signals such as vibration, pressure and currents are processed to produce damage or degradation metrics for the sample population. These metrics are tailored to track general unit degradation or specific degradation modes that developed during the test. The experimental and analytical/numerical procedures are presented along with degradation metrics building and statistical modeling of the data. Examples are given to support the proposed models.